Big Data in Finance
Big data is a popular new catchphrase in the realm of information technology and quantitative methods that refer to the collection and analysis of massive amounts of information. Advances in computing power along with falling prices thereof are making big data projects increasingly more technically feasible and economic. In particular, the advent of cloud computing is putting the cost of big data analysis within the reach of many smaller firms, which now do not need to make significant capital investments in their own computing infrastructure.
A new career category, data science, has sprung up in response to the growth of big data.
Applications Within Finance:
Within finance, particularly within the financial services industry, big data is being utilized in an increasing number of applications, such as:
- Employee monitoring and surveillance
- Predictive models, such as those that may be used by insurance underwriters to set premiums and loan officers to make lending decisions
- Developing algorithms to forecast the direction of financial markets
- Pricing illiquid assets such as real estate
As far back as the 1980s, the founder of Progressive Insurance looked forward to the day when hard data on individual policy holders' driving habits could be collected and analyzed. This would lead to more accurate risk measurement and risk assessment, and thus more precise premium setting. By 2010, the requisite data collection technology had become available, and now over one million customers have agreed to have black boxes installed in their cars that track, for instance, how fast they typically drive and how suddenly they typically brake.
LendUp supplements traditional FICO credit ratings with social network analysis drawn from various other sources, in order to make lending decisions. For example, LendUp is interested in knowing if a potential borrower has changed cell phone numbers frequently, which may indicate a bad risk.
The company also believes that how people interact with their friends online offers strong clues about their riskiness as borrowers. Those who show the strongest and most active social connections and community ties appear to be the best risks. Thus, potential borrowers are asked to make their Facebook accounts available to the firm for analysis.
Credit card giant CapitalOne, meanwhile, became a big player in the 1990s primarily through using advanced data collection and analysis techniques to identify prospects for its cards, stealing a march on many of its more established rivals.
Small Business Lending:
New entrant Kabbage is a thinly-staffed, technology-driven company whose predictive models draw on sources as diverse as social media, eBay and UPS to assess the quality of relationships between potential borrowers and their own customers.
Climate Corporation underwrites crop insurance for farmers. The firm runs huge simulations to predict long-term weather patterns and set premiums.
JPMorgan Chase is using big data analysis to determine acceptable sales prices for homes and commercial properties that have been repossessed as the result of defaulted mortgages.
The idea, according to confidential sources, is to evaluate local economic conditions and property markets to suggest reasonable sales prices before mortgage loans actually go into default. If these suggested sales prices are set accurately, the disruption to the local property market from a default, repossession and sale by the bank theoretically should be minimized. Additionally, the period over which the bank is forced to hold a property prior to making a sale should be minimized.
Meanwhile, Quantfind, a firm that has supplied the CIA with technical expertise to uncover false identities utilized by suspected terrorists, has acknowledged engaging in discussions with JPMorgan Chase over how its technology can be applicable to the credit business, in areas such as credit evaluation and marketing.
Sources: "Data open doors to financial innovation" and "JPMorgan uses counter-terrorism tools to spot fraud among workers," Financial Times, December 14, 2012.